Details
Use case
Handwritten Documents
OCR and specialist agents turn handwritten forms into structured spreadsheets and filing documents.
Problem
Handwritten R&D expense forms arrived as mixed scans, images and multipage PDFs, then had to become normalized schedules and filing documents.
My role
I worked on an anonymized document-automation engagement, connecting OCR, segmentation, agent ordering, spreadsheet output and filing-support artifacts.
Implementation
Use ML/OCR for text recognition and document/section separation, then use agents for contextual ordering, category mapping, normalization into the expected Excel table and generation of filing artifacts.
Engineering choices
Separate visual extraction from business interpretation: OCR/segmentation for handwriting and pages; agents for ordering, category mapping and output formatting.
Stack and domains
- OCR
- ML segmentation
- LLM agents
- Excel/PDF artifacts